federal automotive statistical tool (fast)

Federal Automotive Statistical Tool (FAST) Continuing the Focus on Data Quality

Ron Stewart INL/CON-20-57196:000

• Overview of FAST and how vehicle fleet data is reviewed

• Preliminary review of FY 2019 fleet dataset

• How does it compare to prior year?

• Where are the trouble spots?

• Recommendations

• Discussion

What is FAST?

• Collaborative project funded and managed by

• GSA’s Office of Government-wide Policy

• DOE’s Federal Energy Management Program

• Energy Information Administration

• Collects multiple federal vehicle fleet datasets

• Vehicles & operational data (costs, miles, fuel)

• Fleet budget data

• Fueling infrastructure data

FAST: Why do we care about data quality?

• Based on how the data is used:

• Regulatory & EO compliance determination

• Required publications

• Policy evaluation and development

• Information resource for analysis

• Based on increased interest in Federal vehicle

fleet

• … inside and outside the government

• … particularly since the transition to per-

vehicle reporting

FAST: Why do we care about data quality?

• Implications of questionable data quality

• Inaccurate compliance determination

• Decreased confidence in compliance

determination

• Decreased confidence in dataset for other uses

• Policy development and evaluation

• Information resource for analysis

• Knowing where the potential quality concerns are

is a key part of improving the quality

How is fleet data reviewed?

• As the data is reported:

• Data must pass validation as it is loaded

• 150+ “blocking” checks for basic validity

• Data also screened for reasonability

• 25+ “flagging” checks

• Checks are documented in FAST’s “Vehicle-Level

Data Business Rules Reference”


• After it has been loaded in FAST:

• Pre-defined reports

• User-defined ad hoc queries


• When agency designates submission as

complete:

• FAST team reviews agency submission

• FAST team provides written summary of

noted items

• Agency responds to review summary

• FAST re-opened for corrections identified

by agency

• FAST team assembles review results and

agency responses into dashboard


• FAST team review looks at two levels

• Macro: high-level consistency with past data

• Unexpected shifts in fleet size or makeup?

• Inconsistent shifts in costs, miles, fuel?

• Micro: vehicle-level consistency and trouble spots

• Year-to-year consistency

• Fuel efficiency (e.g., mileage or fuel)

• Use of “placeholders” for vehicle information

• DOE review looks primarily at compliance

• Identifies issues impacting compliance

Preliminary review of FY 2019 fleet data

• Data call officially closed 2019-12-16

• 44 of 50 expected agencies completed on time

• 4 additional agencies complete before 2020-01-01

• 2 small agencies incomplete (~60 vehicles)

• Preliminary review summaries provided to on-time

agencies shortly after completion

• First three agencies had summaries by 2019-11-15

• Significant improvements:

• Agencies are reporting sooner

• Agencies are receiving feedback sooner


• Where do we see trouble spots?

• M1 & M7: Year-to-year consistency of vehicle

reporting

• Missing / inconsistent reporting of vehicles

• Changes in “static” vehicle attributes

• M6: Fuel and mileage at the vehicle level

• M8: Use of “placeholder” values for vehicle

attributes

• Particularly for vehicle fuel and mileage


• Trouble spots: Year-to-year consistency of vehicle

reporting

• Measure 1(a): overall inventory (example)

• Total inventory discrepancies: 15,389 vehicles

• Measure 7(a): Current-year acquisitions also

reported in prior year (flag OW-1.3)

• Total: 2,681 vehicles

• Measure 7(b): Vehicles in current-year inventory

missing from prior year (based on acquisition date;

flag OW-1.4)


Source: FAST (https://fastweb.inl.gov/), 2020-01-13


• Trouble spots: Year-to-year consistency of vehicle

reporting (continued)

• Measure 7(c): Prior-year disposals also present in

current year (flag OW 4.4)


• Measure 7(d): Vehicles in prior year inventory

missing from current year:


• Measure 7(e): Vehicles with year-to-year changes

to “static” attributes (flag A-1.5)




• Trouble spots: Fuel and mileage at the vehicle level

• Measure 6: vehicles with invalid high fuel efficiency

(flag F-4.6)

• Further limited to vehicles w/miles > 1,000

• Total: 40,465 of 581,798 vehicles

• 5 agencies with > 10% of vehicles flagged

• Invalid data skews agency and federal per-

vehicle average fuel efficiency



• Trouble spots: Placeholder values

• Placeholder: vehicle attribute which exactly

matches business rule thresholds for blocking or

flagging data

• Raises question about “real” data

• 72,030 vehicles (of 698,441) reported with one or

more placeholder attributes

• 101,904 blocking placeholders

• 11,963 flagging placeholders


Recommendations

• Maintain perspective:

• Understanding where the problems are is A Good

Thing™

• Improving quality is a (potentially lengthy) process

• Continue the focus on timely reporting

• Use available tools to identify, investigate problem areas

• Flags during data loading process

• Available reports

• Results from reviews (and supporting detail)

• Look for coming changes on the data collection side

FAST Program Contacts

GSA OGP

• James Vogelsinger [email protected]

• Patrick McConnell [email protected]

DOE FEMP

• Kendall Kam [email protected]

• Jay Wrobell [email protected]

EIA CES

• Cynthia Sirk [email protected]

FAST System Team

• Ron Stewart [email protected]

• Michelle Kirby [email protected]

• Tim Raczek [email protected]

• @FASTdevs on Twitter

federal automotive statistical tool (fast)

Documents